Google SAIF (Secure AI Framework)

Google’s framework for building AI systems that are secure and resilient to threats. Provides broader guidance than OWASP LLM Top 10 — covers the full lifecycle of AI system security.

Relevant Risks

Risk	Description
Prompt Injection	Attacks that manipulate the model’s input to cause unintended behavior
Sensitive Data Disclosure	Unintended leakage of sensitive information via the model

These map directly to OWASP LLM01:2025 and LLM02:2025.

⚠️ Placeholder: only two risks captured so far. Ingest the full SAIF risk breakdown to complete this page.

Source: https://saif.google/secure-ai-framework/risks

See Also

• owasp-llm-top-10 — OWASP’s parallel LLM risk framework
• prompt-injection — the shared prompt injection risk
• prompt-engineering-llm-security — prompt engineering & LLM security overview